Info TOP 20 of the World's Most Powerful CPU Cores - IPC/PPC comparison

[Richie Rich]

Added cores:

• A53 - little core used in some low-end smartphones in 8-core config (Snapdragon 450)
• A55 - used as little core in every modern Android SoC
• A72 - "high" end Cortex core used in Snapdragon 625 or Raspberry Pi 4
• A73 - "high" end Cortex core
• A75 - "high" end Cortex core
• Bulldozer - infamous AMD core

Geekbench 5.1 PPC chart 6/23/2020:

Pos​	Man​	CPU​	Core​	Year​	ISA​	GB5 Score​	GHz​	PPC (score/GHz)​	Relative to 9900K​	Relative to Zen3​
1​	Nuvia​	(Est.)​	Phoenix (Est.)​	2021​	ARMv9.0​	2001​	3.00​	667.00​	241.0%​	194.1%​
2​	Apple​	A15 (est.)​	(Est.)​	2021​	ARMv9.0​	1925​	3.00​	641.70​	231.8%​	186.8%​
3​	Apple​	A14 (est.)​	Firestorm​	2020​	ARMv8.6​	1562​	2.80​	558.00​	201.6%​	162.4%​
4​	Apple​	A13​	Lightning​	2019​	ARMv8.4​	1332​	2.65​	502.64​	181.6%​	146.3%​
5​	Apple​	A12​	Vortex​	2018​	ARMv8.3​	1116​	2.53​	441.11​	159.4%​	128.4%​
6​	ARM Cortex​	V1 (est.)​	Zeus​	2020​	ARMv8.6​	1287​	3.00​	428.87​	154.9%​	124.8%​
7​	ARM Cortex​	N2 (est.)​	Perseus​	2021​	ARMv9.0​	1201​	3.00​	400.28​	144.6%​	116.5%​
8​	Apple​	A11​	Monsoon​	2017​	ARMv8.2​	933​	2.39​	390.38​	141.0%​	113.6%​
9​	Intel​	(Est.)​	Golden Cove (Est.)​	2021​	x86-64​	1780​	4.60​	386.98​	139.8%​	112.6%​
10​	ARM Cortex​	X1​	Hera​	2020​	ARMv8.2​	1115​	3.00​	371.69​	134.3%​	108.2%​
11​	AMD​	5900X (Est.)​	Zen 3 (Est.)​	2020​	x86-64​	1683​	4.90​	343.57​	124.1%​	100.0%​
12​	Apple​	A10​	Hurricane​	2016​	ARMv8.1​	770​	2.34​	329.06​	118.9%​	95.8%​
13​	Intel​	1065G7​	Icelake​	2019​	x86-64​	1252​	3.90​	321.03​	116.0%​	93.4%​
14​	ARM Cortex​	A78​	Hercules​	2020​	ARMv8.2​	918​	3.00​	305.93​	110.5%​	89.0%​
15​	Apple​	A9​	Twister​	2015​	ARMv8.0​	564​	1.85​	304.86​	110.1%​	88.7%​
16​	AMD​	3950X​	Zen 2​	2019​	x86-64​	1317​	4.60​	286.30​	103.4%​	83.3%​
17​	ARM Cortex​	A77​	Deimos​	2019​	ARMv8.2​	812​	2.84​	285.92​	103.3%​	83.2%​
18​	Intel​	9900K​	Coffee LakeR​	2018​	x86-64​	1384​	5.00​	276.80​	100.0%​	80.6%​
19​	Intel​	10900K​	Comet Lake​	2020​	x86-64​	1465​	5.30​	276.42​	99.9%​	80.5%​
20​	Intel​	6700K​	Skylake​	2015​	x86-64​	1032​	4.00​	258.00​	93.2%​	75.1%​
21​	ARM Cortex​	A76​	Enyo​	2018​	ARMv8.2​	720​	2.84​	253.52​	91.6%​	73.8%​
22​	Intel​	4770K​	Haswell​	2013​	x86-64​	966​	3.90​	247.69​	89.5%​	72.1%​
23​	AMD​	1800X​	Zen 1​	2017​	x86-64​	935​	3.90​	239.74​	86.6%​	69.8%​
24​	Apple​	A13​	Thunder​	2019​	ARMv8.4​	400​	1.73​	231.25​	83.5%​	67.3%​
25​	Apple​	A8​	Typhoon​	2014​	ARMv8.0​	323​	1.40​	230.71​	83.4%​	67.2%​
26​	Intel​	3770K​	Ivy Bridge​	2012​	x86-64​	764​	3.50​	218.29​	78.9%​	63.5%​
27​	Apple​	A7​	Cyclone​	2013​	ARMv8.0​	270​	1.30​	207.69​	75.0%​	60.5%​
28​	Intel​	2700K​	Sandy Bridge​	2011​	x86-64​	723​	3.50​	206.57​	74.6%​	60.1%​
29​	ARM Cortex​	A75​	Prometheus​	2017​	ARMv8.2​	505​	2.80​	180.36​	65.2%​	52.5%​
30​	ARM Cortex​	A73​	Artemis​	2016​	ARMv8.0​	380​	2.45​	155.10​	56.0%​	45.1%​
31​	ARM Cortex​	A72​	Maya​	2015​	ARMv8.0​	259​	1.80​	143.89​	52.0%​	41.9%​
32​	Intel​	E6600​	Core2​	2006​	x86-64​	338​	2.40​	140.83​	50.9%​	41.0%​
33​	AMD​	FX-8350​	BD​	2011​	x86-64​	566​	4.20​	134.76​	48.7%​	39.2%​
34​	AMD​	Phenom 965 BE​	K10.5​	2006​	x86-64​	496​	3.70​	134.05​	48.4%​	39.0%​
35​	ARM Cortex​	A57 (est.)​	Atlas​	0​	ARMv8.0​	222​	1.80​	123.33​	44.6%​	35.9%​
36​	ARM Cortex​	A15 (est.)​	Eagle​	0​	ARMv7 32-bit​	188​	1.80​	104.65​	37.8%​	30.5%​
37​	AMD​	Athlon 64 X2 3800+​	K8​	2005​	x86-64​	207​	2.00​	103.50​	37.4%​	30.1%​
38​	ARM Cortex​	A17 (est.)​	​	0​	ARMv7 32-bit​	182​	1.80​	100.91​	36.5%​	29.4%​
39​	ARM Cortex​	A55​	Ananke​	2017​	ARMv8.2​	155​	1.60​	96.88​	35.0%​	28.2%​
40​	ARM Cortex​	A53​	Apollo​	2012​	ARMv8.0​	148​	1.80​	82.22​	29.7%​	23.9%​
41​	Intel​	Pentium D​	P4​	2005​	x86-64​	228​	3.40​	67.06​	24.2%​	19.5%​
42​	ARM Cortex​	A7 (est.)​	Kingfisher​	0​	ARMv7 32-bit​	101​	1.80​	56.06​	20.3%​	16.3%​

TOP 10 - Performance Per Area comparison at ISO-clock (PPA/GHz)

Copied from locked thread. They try to avoid people to see this comparison how x86 is so bad.[/B]

Pos​	Man​	CPU​	Core​	Core Area mm2​	Year​	ISA​	SPEC PPA/Ghz​	Relative​
1​	ARM Cortex​	A78​	Hercules​	1.33​	2020​	ARMv8​	9.41​	100.0%​
2​	ARM Cortex​	A77​	Deimos​	1.40​	2019​	ARMv8​	8.36​	88.8%​
3​	ARM Cortex​	A76​	Enyo​	1.20​	2018​	ARMv8​	7.82​	83.1%​
4​	ARM Cortex​	X1​	Hera​	2.11​	2020​	ARMv8​	7.24​	76.9%​
5​	Apple​	A12​	Vortex​	4.03​	2018​	ARMv8​	4.44​	47.2%​
6​	Apple​	A13​	Lightning​	4.53​	2019​	ARMv8​	4.40​	46.7%​
7​	AMD​	3950X​	Zen 2​	3.60​	2019​	x86-64​	3.02​	32.1%​

It's impressive how fast are evolving the generic Cortex cores:

• A72 (2015) which can be found in most SBC has 1/3 of IPC of new Cortex X1 - They trippled IPC in just 5 years.
• A73 and A75 (2017) which is inside majority of Android smart phones today has 1/2 IPC of new Cortex X1 - They doubled IPC in 3 years.

Comparison how x86 vs. Cortex cores:

• A75 (2017) compared to Zen1 (2017) is loosing massive -34% PPC to x86. As expected.
• A77 (2019) compared to Zen2 (2018) closed the gap and is equal in PPC. Surprising. Cortex cores caught x86 cores.
• X1 (2020) is another +30% IPC over A77. Zen3 need to bring 30% IPC jump to stay on par with X1.

Comparison to Apple cores:

• AMD's Zen2 core is slower than Apple's A9 from 2015.... so AMD is 4 years behind Apple
• Intel's Sunny Cove core in Ice Lake is slower than Apple's A10 from 2016... so Intel is 3 years behind Apple
• Cortex A77 core is slower than Apple's A9 from 2015.... but
• New Cortex X1 core is slower than Apple's A11 from 2017 so ARM LLC is 3 years behind Apple and getting closer

GeekBench5.1 comparison from 6/22/2020:

• added Cortex X1 and A78 performance projections from Andrei here
• 2020 awaiting new Apple A14 Firestorm core and Zen3 core

Updated:



EDIT:
Please note to stop endless discussion about PPC frequency scaling: To have fair and clean comparison I will use only the top (high clocked) version from each core as representation for top performance.

 

[IntelUser2000]

These numbers are invalid and you know that. Ryzen 3300X@4.35 GHz cannot have higher score than Ryzen 3950X@4.6 GHz, that's pure nonsense. Ryzen 3300X is probably heavily OCed so using much higher clock than reported 4.35 GHz. That's the problem with Geekbench - you never know the clock and OC status.

It may be invalid for the Ryzen but certainly not for Sunny Cove, as you are ignoring Linux results. And by the way, Linux results are often comparable to Android results.

And while Geekbench pretty much scales linearly, SpecInt2006 doesn't. It typically scales at 80-85%.

 

[Richie Rich]

It may be invalid for the Ryzen but certainly not for Sunny Cove, as you are ignoring Linux results. And by the way, Linux results are often comparable to Android results.

And while Geekbench pretty much scales linearly, SpecInt2006 doesn't. It typically scales at 80-85%.

So you try to tell me that SunnyCove/IceLake has IPC 35% higher than Coffie Lake in 9900K (yours twisted 375 vs. 276 of 9900K). Is it some joke? Because even Intel's optimistic numbers shows 18% increase in average (based on SPECint2006 optimistic numbers).

 

[DrMrLordX]

It's almost as if using Geekbench and SPEC is an invalid way to compare different microarchitectures. Hmm.

 

[Antey]

So you try to tell me that SunnyCove/IceLake has IPC 35% higher than Coffie Lake in 9900K (yours twisted 375 vs. 276 of 9900K). Is it some joke? Because even Intel's optimistic numbers shows 18% increase in average (based on SPECint2006 optimistic numbers).

so sunny cove IPC is what you want or believe it to be?

 

[Richie Rich]

It's almost as if using Geekbench and SPEC is an invalid way to compare different microarchitectures. Hmm.

Attempt to start flame? I'm still waiting that you will prove this faulty statement. Still waiting you to show us some CPU that excels in SPEC or GeekBench but has terrible performance in real world application. Waiting already more than one year and no single proof from you and others.

I posted here Blender results: RPI4 (A72) vs. my Ryzen 3700X. And it matched the GeekBench results.

I thought that you will stop spreading this invalid statements with Fujitsu supercomputer based on ARM and with Apple's transition their bilion bussiness to ARM architecture as another proof. At least 90% of people stopped laughing at Apple's A13 performance because Apple would never migrate Mac Pro workstations to ARM if there is no performance.

so sunny cove IPC is what you want or believe it to be?

I want valid numbers. If somebody tells me that Ice Lake has 35% higher IPC than 9900K, obviously it's fake number. I know you try hard to discredit this PPC/IPC chart. But in the end you just discredit yourself as time proves this PPC table is correct (Fukagu, Apple migration to ARM, server CPU Graviton2 results on Phoronix). And first benchmarks and real world applications of new ARM cores like A14 and A78/X1 will hammer last nail into naysayer's coffin.

 

[Antey]

I want valid numbers

the problem is you don't have any methodology, you just believe something is valid because it matches what you have read about that microarchitecture IPC, you will only take the results that match what you think is skylake, zen 2 or sunny cove IPC.

i could make my own chart with very very different results with the same source as yours, some random geekbench results which i think are 'valid'. in fact, this could be a clownfest with multiple charts. call it reductio ad absordum, if you like.

if you want to prove that apple SPEC and geekbench results are incredible, yes, they are. otherwise, your methodology is flawed. and the more i think about it the more i think i will follow the rest and ignore this post.

 

[Mr Evil]

It's an interesting comparison, and although I know it's a little off-topic, I was inspired to do my own similar comparison of power-per-litre for internal combustion engines.

Pos	Man	Model	Engine Type	Year	BHP	Litres	PPL	Relative
1	Mazda	RX-8	Wankel	2003	238	1.3	183	100%
2	Bugatti	Veyron	Quad turbo W16	2005	987	8.0	123	67%
3	Maersk	Emma Maersk	2-stroke turbo diesel	2006	107390	25340	4	2%

Conclusion:
Wankel engines are seriously impressive, and if Mazda can just scale them up a bit they could revolutionize the shipping industry, with a 46x increase in power output for the same engine size.

 

[DrMrLordX]

Still waiting you to show us some CPU that excels in SPEC or GeekBench but has terrible performance in real world application.

Yeah, because nobody cheats in SPEC with compiler tricks.

 

[Richie Rich]

the problem is you don't have any methodology, you just believe something is valid because it matches what you have read about that microarchitecture IPC, you will only take the results that match what you think is skylake, zen 2 or sunny cove IPC.

i could make my own chart with very very different results with the same source as yours, some random geekbench results which i think are 'valid'. in fact, this could be a clownfest with multiple charts. call it reductio ad absordum, if you like.

if you want to prove that apple SPEC and geekbench results are incredible, yes, they are. otherwise, your methodology is flawed. and the more i think about it the more i think i will follow the rest and ignore this post.

SPEC results are taken from AnandTech articles and microarchitecture breakdowns from Andrei Frumusanu. Are you trying to say that official AnandTech benchmark measurements are flawed? Are you joking?

Arm's New Cortex-A78 and Cortex-X1 Microarchitectures: An Efficiency and Performance Divergence

www.anandtech.com

It's an interesting comparison, and although I know it's a little off-topic, I was inspired to do my own similar comparison of power-per-litre for internal combustion engines.

Pos	Man	Model	Engine Type	Year	BHP	Litres	PPL	Relative
1	Mazda	RX-8	Wankel	2003	238	1.3	183	100%
2	Bugatti	Veyron	Quad turbo W16	2016	987	8.0	123	67%
3	Maersk	Emma Maersk	2-stroke turbo diesel	2006	107390	25340	4	2%

Conclusion:
Wankel engines are seriously impressive, and if Mazda can just scale them up a bit they could revolutionize the shipping industry, with a 46x increase in power output for the same engine size.

Yeah, it's off-topic and total non-sense. You can add rocket engine Merlin 1D with its 2GW power and revolutionize everything. Using your anology: Apple A13 is powerfull like rocket engine while sipping power like EV Tesla Model 3. I hope you understand now why Apple moved to ARM.

 

[Richie Rich]

Yeah, because nobody cheats in SPEC with compiler tricks.

You're saying AnandTech cheats in its own SPEC results? Unless you have proof you shouldn't use such a serious accusation.

 

[DrMrLordX]

You're saying AnandTech cheats in its own SPEC results?

Oh please.

https://www.reddit.com/r/hardware/comments/6if8vw/spec_updates_its_cpu_benchmark_for_the_first_time/

Or if you just want the article without reddit commentary:

https://techreport.com/news/32126/spec-updates-its-cpu-benchmark-for-the-first-time-in-11-years/

I don't know if ICC cheats in SPEC2017 now, too, though it wouldn't surprise me.

Honest question: do you even understand the implications of using SPEC2006 and GB5 numbers for IPC reference data? Can you think of anything wrong with these benchmarks? It would really help you if you could sort some of this out yourself instead of relying on others to spoonfeed you, since you don't seem to want anyone to contradict you anyway.

There's a lot that I didn't know about SPEC even until recently. You can learn too.

 

[Hitman928]

I posted here Blender results: RPI4 (A72) vs. my Ryzen 3700X. And it matched the GeekBench results.

No it didn't, your A72 significantly under performed compared to its GB result and you tried to blame it on a lack of memory bandwidth which wasn't very likely as Ryzen showed only a minor performance degradation with the same memory bandwidth as Rpi4.

 

[Mr Evil]

You can add rocket engine Merlin 1D with its 2GW power and revolutionize everything.

Don't be silly - that's an external combustion engine. You can only compare like with like.

 

[Richie Rich]

Oh please.

https://www.reddit.com/r/hardware/comments/6if8vw/spec_updates_its_cpu_benchmark_for_the_first_time/

Or if you just want the article without reddit commentary:

https://techreport.com/news/32126/spec-updates-its-cpu-benchmark-for-the-first-time-in-11-years/

I don't know if ICC cheats in SPEC2017 now, too, though it wouldn't surprise me.

Honest question: do you even understand the implications of using SPEC2006 and GB5 numbers for IPC reference data? Can you think of anything wrong with these benchmarks? It would really help you if you could sort some of this out yourself instead of relying on others to spoonfeed you, since you don't seem to want anyone to contradict you anyway.

There's a lot that I didn't know about SPEC even until recently. You can learn too.

So you accuse AnandTech cheated the tests since 2019 A13 release?
Because Intel cheated in 2017 with his compiler in libquantum subtest?
That's the reason why A13 has higher PPC of massive 83% over Zen2? (Even it has nothing to do with Intel's compiler?)
Are you sure Andrei was using Intel's x86 compiler for Apple's A13 ARM CPU?

 

[Thunder 57]

@Richie Rich

I suggest you read this article.

From said article:

At the same time, however, Apple isn’t shifting to ARM in a year, the way it did with x86 chips. Instead, Apple hopes to be done within two years. One way to read this decision is to see it as a reflection of Apple’s long-term focus on mobile. Scaling a 3.9W iPhone chip into a 15-25W laptop form factor is much easier than scaling it into a 250W TDP desktop CPU socket with all the attendant chipset development required to support things like PCIe 4.0 and standard DDR4 / DDR5 (depending on launch window).

You play with numbers of "Start flame"? You gave me a joke of a response to an A14 laptop clobbering a 3950X. You don't seem to want to reason or discuss anything. In your mind, everything you say is fact. That is why so many have trouble with you.

 

[lobz]

So you accuse AnandTech cheated the tests since 2019 A13 release?
Because Intel cheated in 2017 with his compiler in libquantum subtest?
That's the reason why A13 has higher PPC of massive 83% over Zen2? (Even it has nothing to do with Intel's compiler?)
Are you sure Andrei was using Intel's x86 compiler for Apple's A13 ARM CPU?

In the middle of this great laughter, try to answer Hitman's post #87

 

[gdansk]

Can't wait for the ARM MacBook whenever that launches. I'll soon be able to compare this to my 9750H MacBook for the code I actually run.
But gosh, Mac OS 11 looks like a serious downgrade from Catalina which was a buggy mess.

 

[IntelUser2000]

So you try to tell me that SunnyCove/IceLake has IPC 35% higher than Coffie Lake in 9900K (yours twisted 375 vs. 276 of 9900K). Is it some joke? Because even Intel's optimistic numbers shows 18% increase in average (based on SPECint2006 optimistic numbers).

I sit down and analyze results first before I post. Do you?

9900K's result will be higher in Linux/Android too. Desktops can be overclocked much readily making peak result difficult to isolate.

18% is not optimistic. It's average.

https://images.anandtech.com/doci/14514/Blueprint%20Series_May%2016-2019_COMBINED%20FINAL_AnandTech%20%282%29-page-029.jpg

 

[Antey]

Ok, i made my own WebXPRT3 PPC chart, anandtech uses it so is 100% valid (yes, i had to say it)

WebXPRT3 PPC chart 6/26/2020:

Pos​	Man​	CPU​	Core​	Year​	ISA​	WebXPRT3 Score​	GHz​	PPC (score/GHz)​	Relative​	Relative​
1​	Apple​	A13​	Lightning​	2019​	ARMv8​	187​	2,65​	70,6​	100,00 %​	103,42 %​
2​	AMD​	3950X​	Zen 2​	2019​	x86-64​	317​	4,6​	68,9​	97,66 %​	100,99 %​
3​	Intel​	9900K​	Skylake​	2018​	x86-64​	348​	5,1​	68,2​	96,70 %​	100,00 %​
4​	Apple​	A12​	Vortex​	2018​	ARMv8​	163​	2,53​	64,4​	90,80 %​	94,42 %​
5​	AMD​	1800X​	Zen 1​	2017​	x86-64​	237​	4​	59,3​	83,96 %​	86,83 %​
6​	Apple​	A11​	Monsoon​	2017​	ARMv8​	137​	2,39​	57,3​	80,81 %​	84,01 %​
7​	ARM Cortex​	A77​	Deimos​	2019​	ARMv8​	120​	2,84​	42,3​	59,88 %​	61,92 %​
8​	Apple​	A10​	Hurricane​	2016​	ARMv8​	82​	2,34​	35,0​	49,36 %​	51,36 %​
9​	ARM Cortex​	A76​	Enyo​	2018​	ARMv8​	97​	2,84​	34,2​	48,40 %​	50,05 %​

PT - WebXPRT results

Principled Technologies Durham, NC, Industry-leading technology assessment, fact-based marketing services, hands-on testing of technology products, deliver the facts, testing reports and videos

www.principledtechnologies.com

EDIT: This is not serious.

 

[myocardia]

My Pentium D made the list I'm happy see no reason to upgrade.

I know, right? I have no idea why you are using that slow 3700X, when you could be using a blazing-fast Pentium D!

 

[DrMrLordX]

So you accuse AnandTech cheated the tests since 2019 A13 release?

Part of the problem with these attempts at conversation is that you put words in people's mouths thusly. People have been cheating at SPEC in general, and usually its the compiler vendors, NOT the individuals who conduct the benchmarks themselves. So no, Anandtech specifically isn't cheating their SPEC numbers. But any SPEC2006 results they (or anyone else) have ever published after compiling with ICC for x86 targets would have inadvertently cheated libquantum, at a minimum. Which should tell you that other cheats are also possible . . . they might just not have been caught yet. SPEC, as a benchmark, is not the most reliable thing in the world, and SPEC2006 in particular is rather old now.

Now that you've attempted to beat that dead horse, would you care to consider all the other problems with SPEC2006 and GB5? Before others have to tell you?

 

[Doug S]

I think you missed his point. The Apple A series is highly optimized for single threaded workloads, unlike the 3950x and other high end x86-64 CPUs which are highly optimized for multithreaded workloads as well.

Sorry, there's no such thing as optimizing a core for single threaded workloads. You do realize that the A series has 2 or 4 big cores plus 4 little cores, right? Every CPU is multicore these days. Making a CPU perform well in 4 way vs 16 way or 64 way has almost no effect on its single threaded performance. More cores will add a little bit of latency to main memory but the overall performance hit is low single digit in magnitude at most.

All you guys who are holding onto this belief that Apple's SoCs will somehow be exposed when A14 based Macs ship in six months are going to be doing a lot of backtracking and crow eating.

 

[Richie Rich]

Part of the problem with these attempts at conversation is that you put words in people's mouths thusly. People have been cheating at SPEC in general, and usually its the compiler vendors, NOT the individuals who conduct the benchmarks themselves. So no, Anandtech specifically isn't cheating their SPEC numbers. But any SPEC2006 results they (or anyone else) have ever published after compiling with ICC for x86 targets would have inadvertently cheated libquantum, at a minimum. Which should tell you that other cheats are also possible . . . they might just not have been caught yet. SPEC, as a benchmark, is not the most reliable thing in the world, and SPEC2006 in particular is rather old now.

Now that you've attempted to beat that dead horse, would you care to consider all the other problems with SPEC2006 and GB5? Before others have to tell you?

Andrei uses GCC and LLVM compilers:

Amazon's Arm-based Graviton2 Against AMD and Intel: Comparing Cloud Compute

www.anandtech.com

Another your fake-news I destroyed. It was pretty long list of fake news:

• Apple's are designed only for short bursts, it cannot handle sustained load
• Apple is using ASIC circuits to increase IPC
• Apple score is caused by cheating via compiler (even it's impossible to cheat all 40 subtests for SPEC and Geekbench when using GCC and LLVM)
• ARM cores cannot scale up into many-core server CPUs (Phoronix showed Graviton2 is even better than A76 SPEC results, so SPEC and GB results are valid)

All these pearls are describing pretty nicely how desperate you are from Apple's performance. I wonder what will be the next performance denial excuse? Green Martians hidden in the chip?

And time to ARM MacBook release is ticking out fast. The truth who is right is coming soon

 

[DrMrLordX]

Andrei uses GCC and LLVM compilers:

And you've missed my point entirely. But hey, thanks for trying.

All you guys who are holding onto this belief that Apple's SoCs will somehow be exposed when A14 based Macs ship in six months are going to be doing a lot of backtracking and crow eating.

Nobody's eating crow. The problem is that some people are using . . . questionable techniques to try to compare mobile SoC cores to desktop and server cores.

Apple's A-series chips may actually be faster than anyone - including Richie Rich - suspects. Or they could be slower. When you use complete horse manure to try to create a realistic picture of relative performance, you just don't know. Hell I can't even get Geekbench 5 to significantly tax my 3900x in GB5. Is it even using AVX or AVX2? The only tests that seemsto push it at all are Structure from Motion and Ray Tracing.

To contrast GB5 with CBR20 (4.4 GHz, 1.344v-1.38v)

CBR20:

ST: ~49W package power
MT: ~162W package power

GB5:

ST: ~42W average, peaked at 46W in Structure from Motion
MT: All over the place, so average seems pointless, but it was ~91W. Ray tracing seemed to push power up to around 142-147W, while Structure from Motion hit the 130s.

Pretty sure Primate Labs claims to use AVX, but . . .

 

[Carfax83]

Sorry, there's no such thing as optimizing a core for single threaded workloads.

When you quoted me I didn't use the word "core." I said Apple A series, meaning the entire chip. And while you're probably right that you can't really optimize a core for single threaded workloads, you can certainly optimize a chip/CPU.

You do realize that the A series has 2 or 4 big cores plus 4 little cores, right? Every CPU is multicore these days. Making a CPU perform well in 4 way vs 16 way or 64 way has almost no effect on its single threaded performance. More cores will add a little bit of latency to main memory but the overall performance hit is low single digit in magnitude at most.

Yes I'm aware, and I disagree. First off, a disclaimer. I'm not a chip designer or anything, so I'm just going off of my observation. From what I've seen, cache hierarchy is perhaps the number one factor affecting scalability. Then after that, it's interconnect. Or perhaps they're reversed? I'm sure someone will correct me if I'm wrong.

Anyway, Intel, AMD and IBM's high core count CPUs all have a few things in common; small L1 and L2 caches, and absolutely humongous L3 caches. This is in sharp contrast to the Apple A series which has a huge L1 and L2 cache, with no L3 cache.

For the workloads that Intel, AMD and IBM have designed their CPUs, that cache hierarchy makes sense because with so many cores on a single die, the much slower L3 acts as a common cache between them so that some or all of the cores can share the multithreaded workloads without taking a long trip to the main memory.

Juxtaposed to the Apple A series which has a very large low latency L2 cache which is shared between two of the big cores, because the workloads on a mobile device tend to be single threaded for the most part. I would imagine that for the most part, a single core will usually have all 8MB of L2 cache to itself.

This helps to explain why the A series is so fast in single threaded workloads, because memory latency is a huge obstacle for CPU performance. And if Apple ever scaled the A series to say 64 cores a la Epyc, they would certainly have to abandon their current cache hierarchy in favor of the one that is used by AMD, Intel and IBM.

All you guys who are holding onto this belief that Apple's SoCs will somehow be exposed when A14 based Macs ship in six months are going to be doing a lot of backtracking and crow eating.

I guess we'll see. I'm looking forward to seeing Apple duke it out with Intel and AMD in anything but Geekbench and Spec2006.